System gets hanged and f measure difference

Question

System gets hanged and f measure difference

tuhinkm opened this issue 5 years ago · comments

tuhinkm commented 5 years ago

Hi,
I am having the following issues/doubts:

My system gets hanged when i try to run the saliency part of the code, for both test.py and predict.py. I use the pretrained model shared. Can you please help.
The paper states the F measure to be around 91% on DSS. However, the DSS method itself has reported a 92% F measure. Can you please explain the reason for the difference.

Wu Huikai · Answer 1 · Mon Mar 04 2019 10:45:26 GMT+0800 (China Standard Time)

Please give me more information about your problem.
We couldn't reproduce the reported F measure (92% ) with Pytorch :(

tuhinkm · Answer 2 · Mon Mar 04 2019 14:54:08 GMT+0800 (China Standard Time)

Thanks for the promt reply.

I get the one line print of the parser_args outputs. Then the system hangs for testing. Not sure why.
Can you please share your saliency (saliency maps) results reported in google drive, with and without Deep Guided Filter. Will be of great help for comparison.

Wu Huikai · Answer 3 · Mon Mar 04 2019 20:33:39 GMT+0800 (China Standard Time)

Without much info, I couldn't reproduce your problem. I suggest using PyCharm for debugging.
Currently, I'm busy with ICCV. I can release the results after then.

Wu Huikai · Answer 4 · Mon Mar 25 2019 16:34:52 GMT+0800 (China Standard Time)

The saliency maps are in here, where dss_dgf.zip is with DGF and dss.zip is without DGF.
{id}.png is the ground truth, {id}_sal.png is the saliency map and {id}_sal_{thres}.png is the result after applying a threshold.
test.txt is the list of test images and test_result.txt is the performance.